PREDICTING CLASS OF DEGREE OF UNIVERSITY STUDENTS USING ENSEMBLE MACHINE LEARNING

Machine learning (ML) models are increasingly employed in educational settings to uncover patterns that can improve decision-making and bring improvements to the educational system. Although significant progress has been made in predicting academic achievement, there are limited exploration in combining multiple algorithms to enhance prediction accuracy. This research evaluates the effectiveness of six traditional machine learning algorithms; Decision Trees (DT), Random Forest (RF), Multilayer Perceptron (MLP), Support Vector Machines (SVM), Naïve Bayes (NB) and K-Nearest Neighbors (KNN) in predicting students' class of degree, using 2101 instances of the dataset of Mathematic, Statistics and Computer Science students, University of Abuja, Nigeria. Random Forest (RF) achieved the highest accuracy of 88.89%, followed by Decision Tree (82.52%) and Support Vector Machines (SVM) (81.55%), while Multilayer Perceptron and K-Nearest Neighbors (KNN) showed lower effectiveness. The research applied advanced ensemble methods; the Stacked Generalization, Bagging and Boosting to the traditional ML to enhance accuracy, which significantly improved the performance of the models. Stacked Generalization achieved an impressive accuracy of 96.95%, and Bagged Random Forest reached 93.54% while the Bagged Decision Tree achieved 93.02% accuracy.